FittedBondDiscountCurve with TBills

[Steve Townsend]

I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight settings for duration < 1.0 seem too lage, skewing the fitting undesirably.  What is a sensible choice of weight formula across the maturity spectrum when maturities < 1Y are needed on the curve?  Seems like preserving the existing logic for >= 1Y is fine but a different rule is needed for < 1Y.

------------------------------------------------------------------------------

_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Andres Hernandez]

We weight them by liquidity, and as a rough measure we use the mid price divided by the bid-ask spread (we actually use relative weights, so this one is divided by the sum of all). Then again, as objective function we don't use the squared sum of errors, but rather something close to it. If the quoting error is actually within the bid-ask spread, then the error contributes only marginally (divided by a constant, normally 10). The intention is to allow for "free" movement of the curve within the bid-ask spread but to penalize any result that would put the value outside of it.


Mit freundlichen Grüßen / Kind regards

Dr. Andres Hernandez

Senior Financial Engineer
Business Analytics
Risk Analytics

Phone:	[phone]	IBM Deutschland
Mobile:	[phone]	Wilhelm-Fay-Str. 30-34
E-Mail:	[hidden email]	65936 Frankfurt am Main
		Germany
IBM Deutschland GmbH / Vorsitzender des Aufsichtsrats: Martin Jetter Geschäftsführung: Martina Koederitz (Vorsitzende), Susanne Peter, Norbert Janzen, Dr. Christian Keller, Ivo Koerner, Markus Koerner Sitz der Gesellschaft: Ehningen / Registergericht: Amtsgericht Stuttgart, HRB 14562 / WEEE-Reg.-Nr. DE 99369940

Steve Townsend ---26/10/2015 19:31:50---I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight setting

From: Steve Townsend <[hidden email]>
To: [hidden email]
Date: 26/10/2015 19:31
Subject: [Quantlib-users] FittedBondDiscountCurve with TBills

---

I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight settings for duration < 1.0 seem too lage, skewing the fitting undesirably.  What is a sensible choice of weight formula across the maturity spectrum when maturities < 1Y are needed on the curve?  Seems like preserving the existing logic for >= 1Y is fine but a different rule is needed for < 1Y.
------------------------------------------------------------------------------
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users



------------------------------------------------------------------------------

_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Luigi Ballabio]

Also, Andres contributed a pull request already that makes it possible to specify weights during fitting. We're smoothing out a few glitches, but I'm aiming at including it in next release.

Luigi

On Tue, Oct 27, 2015 at 8:51 AM Andres Hernandez <[hidden email]> wrote:

We weight them by liquidity, and as a rough measure we use the mid price divided by the bid-ask spread (we actually use relative weights, so this one is divided by the sum of all). Then again, as objective function we don't use the squared sum of errors, but rather something close to it. If the quoting error is actually within the bid-ask spread, then the error contributes only marginally (divided by a constant, normally 10). The intention is to allow for "free" movement of the curve within the bid-ask spread but to penalize any result that would put the value outside of it.


Mit freundlichen Grüßen / Kind regards

Dr. Andres Hernandez

Senior Financial Engineer
Business Analytics
Risk Analytics

Phone:	[phone]	IBM Deutschland
Mobile:	[phone]	Wilhelm-Fay-Str. 30-34
E-Mail:	[hidden email]	65936 Frankfurt am Main
		Germany
IBM Deutschland GmbH / Vorsitzender des Aufsichtsrats: Martin Jetter Geschäftsführung: Martina Koederitz (Vorsitzende), Susanne Peter, Norbert Janzen, Dr. Christian Keller, Ivo Koerner, Markus Koerner Sitz der Gesellschaft: Ehningen / Registergericht: Amtsgericht Stuttgart, HRB 14562 / WEEE-Reg.-Nr. DE 99369940

Steve Townsend ---26/10/2015 19:31:50---I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight setting

From: Steve Townsend <[hidden email]>
To: [hidden email]
Date: 26/10/2015 19:31
Subject: [Quantlib-users] FittedBondDiscountCurve with TBills

---

I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight settings for duration < 1.0 seem too lage, skewing the fitting undesirably.  What is a sensible choice of weight formula across the maturity spectrum when maturities < 1Y are needed on the curve?  Seems like preserving the existing logic for >= 1Y is fine but a different rule is needed for < 1Y.
------------------------------------------------------------------------------
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

------------------------------------------------------------------------------
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

--

<http://leanpub.com/implementingquantlib>
<http://implementingquantlib.com>
<http://twitter.com/lballabio>

------------------------------------------------------------------------------

_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

[Steve Townsend]

Many thanks for the info.  I'll keep an eye out for that update in the repo.

On Tue, Oct 27, 2015 at 5:38 AM, Luigi Ballabio <[hidden email]> wrote:

Also, Andres contributed a pull request already that makes it possible to specify weights during fitting. We're smoothing out a few glitches, but I'm aiming at including it in next release.

Luigi

On Tue, Oct 27, 2015 at 8:51 AM Andres Hernandez <[hidden email]> wrote:

We weight them by liquidity, and as a rough measure we use the mid price divided by the bid-ask spread (we actually use relative weights, so this one is divided by the sum of all). Then again, as objective function we don't use the squared sum of errors, but rather something close to it. If the quoting error is actually within the bid-ask spread, then the error contributes only marginally (divided by a constant, normally 10). The intention is to allow for "free" movement of the curve within the bid-ask spread but to penalize any result that would put the value outside of it.


Mit freundlichen Grüßen / Kind regards

Dr. Andres Hernandez

Senior Financial Engineer
Business Analytics
Risk Analytics

Phone:	<a href="tel:%2B49-69-6645-1351" value="[phone]" target="_blank">[phone]	IBM Deutschland
Mobile:	<a href="tel:%2B49-163-609-3280" value="[phone]" target="_blank">[phone]	Wilhelm-Fay-Str. 30-34
E-Mail:	[hidden email]	65936 Frankfurt am Main
		Germany
IBM Deutschland GmbH / Vorsitzender des Aufsichtsrats: Martin Jetter Geschäftsführung: Martina Koederitz (Vorsitzende), Susanne Peter, Norbert Janzen, Dr. Christian Keller, Ivo Koerner, Markus Koerner Sitz der Gesellschaft: Ehningen / Registergericht: Amtsgericht Stuttgart, HRB 14562 / WEEE-Reg.-Nr. DE 99369940

Steve Townsend ---26/10/2015 19:31:50---I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight setting

From: Steve Townsend <[hidden email]>
To: [hidden email]
Date: 26/10/2015 19:31
Subject: [Quantlib-users] FittedBondDiscountCurve with TBills

---

I am trying to extend the Bond Curve to support T-Bills on the short end, The default weight settings for duration < 1.0 seem too lage, skewing the fitting undesirably.  What is a sensible choice of weight formula across the maturity spectrum when maturities < 1Y are needed on the curve?  Seems like preserving the existing logic for >= 1Y is fine but a different rule is needed for < 1Y.
------------------------------------------------------------------------------
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

------------------------------------------------------------------------------
_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users

--

<http://leanpub.com/implementingquantlib>
<http://implementingquantlib.com>
<http://twitter.com/lballabio>

------------------------------------------------------------------------------

_______________________________________________
QuantLib-users mailing list
[hidden email]
https://lists.sourceforge.net/lists/listinfo/quantlib-users